1EE244 Project Proposal:EE244 Project Proposal:Process Aware Timing AnalysisProcess Aware Timing AnalysisWojtek Poppe ,Zheng Guo, and ChungWojtek Poppe ,Zheng Guo, and Chung--Hsun LinHsun LinEECS Department, U. C. BerkeleyEECS Department, U. C. BerkeleyGood points: Title slide – we know who is doing the project!!!EE244 Project Proposal 2MotivationMotivationProcess variations have a significant effect on the circuit performance and yield. This problem will only get worse in the future technology generations.Statistical distribution of transistor threshold voltages do not capture enough processing details.Random threshold voltage variations is not well characterized.There is big gap between circuit designers and processing engineers.Good points: Timely, not project dependentWeaknesses in other projects: Tied to specific projects not general industry/technology trends2EE244 Project Proposal 3Problem DefinitionProblem DefinitionHow to effectively communicate processing knowledge to circuit designersAerial simulator can predict CD distribution based on layoutSignificant amounts of variations can be predicted by moving within the your process windowHow to use processing knowledge to design more robust circuits with minimum trade-offIdentify problem areas with aerial image and circuit simulationDiscriminately apply robust circuit techniques locallyIdentify what portion of random threshold voltage variation can be attributed to predictable systematic CD variationA little general, needs to be focused furtherEE244 Project Proposal 4Literature SurveyLiterature SurveyJie Yang, Capodieci L, Sylvester D. Advanced timing analysis based on post-OPC extraction of critical dimensions. IEEE. 2005, pp.359-64. Piscataway, NJ, USAThis paper defines an aerial image simulator to static timing analyzer flow that we are going to use in our project.Orshansky M, Milor L, Pinhong Chen, Keutzer K, Chenming Hu. Impact of spatial intrachipgate length variability on the performance of high-speed digital circuits., vol.21, no.5, May 2002, pp.544-53. Publisher: IEEE, USA. One of the first papers to propose looking at transistors in a statistical fashion based on emperical Lgate data.Borkar S. Circuit techniques for subthreshold leakage avoidance, control and tolerance. IEEE. 2005, pp.421-4. Piscataway, NJ, USA.Friedberg P, Cao Y, Cain J, Wang R, Rabaey J, Spanos C. Modeling within-die spatial correlation effects for process-design co-optimization. Proceedings. 6th International Symposium on Quality Electronic Design. IEEE Comput. Soc. 2005, pp.516-21. Los Alamitos, CA, USA. Paul used emperical data from Jason Cain's thesis that characterized intrachip, intra wafer, and inter wafer CD variation.These papers' statistics are skewed as CD data is extracted from perfect gratings which do not appear on actual chip layouts. The paper finds mostly systematic CD variation, which should be addressed by processing engineers since it is mostly layout/proximity independent. Designers should address CD variation effects that can be mitigated by changes in the design.Good points: real bibliography, explains relevance of papers, also course professor is referenced – always a good idea ☺☺☺☺3EE244 Project Proposal 5Literature SurveyLiterature SurveyJ. A. G. Jess, K. Kalafala, S. R. Naidu, R. H. J. M. Otten, and C. Visweswariah. Statistical timing for parametric yield prediction of digital integrated circuits. Design Automation Conference (DAC), Anaheim, CA, pages 932--937, June 2003.Nice approach for statistical timing analysis based on Einstimer’s static timer. However, like most work on statistical timing, it doest not provide a good link between processing and actual circuit implementation.C. Visweswariah, K. Ravindran, K. Kalafala, S. G. Walker, and S. Narayan. First-order incremental block-based statistical timing analysis. Design Automation Conference (DAC), San Diego, CA, pages 331--336, June 2004.Velenis, D.; Sundaresha, R.; Friedman, E.G. Buffer sizing for delay uncertainty induced by process variations. Electronics, Circuits and Systems, 2004. ICECS 2004. Proceedings of the 2004 11th IEEE International Conference on 13-15 Dec. 2004 Page(s):415 – 418May be helpful for robust circuit techniques.Agarwal, A.; Blaauw, D.; Zolotov, V.;” Statistical timing analysis for intra-die process variations with spatial correlations,”Computer Aided Design, 2003. ICCAD-2003. International Conference on9-13 Nov. 2003 Page(s):900 - 907 Nice reference for new statistical timing analysis algorithm. Yu Cao; Clark, L.T.; “Mapping statistical process variations toward circuit performance variability: an analytical modeling approach,” Design Automation Conference, 2005. Proceedings. 42nd13-17 June 2005 Page(s):658 – 663Using analytical model to calculate the impact of process variation on the delay. Nice for comparison.Venkatraman, V.; Burleson, W.;“Impact of process variations on multi-level signaling for on-chip interconnects,” VLSI Design, 2005. 18th International Conference on2005 Page(s):362 - 367 EE244 Project Proposal 6Proposed ApproachProposed ApproachAutomate aerial image simulation to take GDS file and output list of gate CDs (Mentor Graphics CalibreWorkbench)Link aerial image simulation to HSPICEBuild delay look-up table reference by processing conditions (dose and focus) within process window (HSPICE)Static timing analysis using look-up table (Primetime?)Evaluate novel robust circuit design techniquesExtra Credits: investigate circuits for project validationA little general, needs to be focused further4EE244 Project Proposal 7TimelineTimelineWeek 1-3: Infrastructure, research of robust circuit techniquesWeek 3-6: build separate simulation modulesWeek 7-9: link modules together and evaluate robust circuit design techniquesEE244 Project Proposal 8Problem DefinitionProblem Definition--Loo, WangLoo, WangSpaceWire is a general purpose interconnection network for spacecrafts.Given RTL verilog for SpaceWire’s CODEC (i.e. its transmitter and receiver modules) the following specifications need to be verified:Parity bit errors, credit errors, empty packets, escape errors, and data link disconnects must be detectedUpon error detection, the CODEC must stop current operations andenter reset state.After a reset, CODEC must follow an “exchange of silence”protocol to
View Full Document